JPH08171569A - Document retrieval device - Google Patents

Abstract

PURPOSE: To perform retrieval with a little retrieval omission and a little retrieval noise by selecting a word which is closed to the retrieval intention OF a user when a key word for retrieval is expanded. CONSTITUTION: A dictionary means 14 holds information for finding words relating to one word and the categories of the words. A judging means 15 judges whether or not the key word for retrieval is classified into plural categories by referring to the dictionary means 14. When the key word is classified into plural categories, a specifying means 16 takes >=1 words classified into them out of the categories and displays them to make the user specify one of them. When the key word is classified into one category, nothing is done. A key word expanding means 13 finds the category of the key word for retrieval by referring to the dictionary means 14 to take words belonging to the category out. A retrieval means 17 retrieves document data with the key word for retrieval and the words taken out by the key words expanding means 13.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for expanding a keyword in a given search key by a thesaurus or the like in a document search, and more particularly to an apparatus for expanding a search key into a plurality of categories. Regarding processing for classified words.

[0002]

2. Description of the Related Art Conventionally, in order to eliminate omission of a search and to support a search for a document according to a search intention, generally, a key extracted from a question sentence or a search expression by a keyword extraction means is
The keyword expansion means refers to the thesaurus dictionary to automatically expand synonyms and upper / lower words, and searches using the expanded words as search keys. For example, JP-A-63-168721 and JP-A-3-13
Japanese Patent No. 0873 proposes to expand a search key as a synonym (related word) and perform a search as the search key. Further, Japanese Patent Application Laid-Open No. 4-10062 proposes to extract a superordinate / subordinate relationship, automatically develop a key, and perform retrieval as a retrieval key.

[0003]

However, in these cases, when one word is classified into a plurality of categories,
If it is automatically expanded, a word that does not match the search intention is expanded as a search key and unnecessary documents are searched for, resulting in a problem of search noise that the search efficiency becomes poor. For example, if you want to search for "teacher (school)", you can search for "teacher, teacher, professor, doctor, doctor, teacher,
Since it will be expanded to "Guru", even words that do not match the search intention will be used as search keys.

On the other hand, Japanese Laid-Open Patent Publication No. 4-145578 proposes a method of displaying upper and lower nodes in a tree structure so that the user can select a word that does not match the search intention and use it as a search key. Has been done. But,
In this case, you need a dictionary with a specific structure, and even if you can overcome it and display it, it is good if it is represented by a simple tree structure, but multiple categories (concepts)
In the case represented by, there is a problem that the search key cannot be easily specified because the number of them is large and the relation structure becomes complicated.

According to the present invention, when the related keyword is expanded by the keyword expanding means, only the words that match the user's search intention can be selected, and the omission of search and the occurrence of search noise can be prevented. The purpose is to be able to.

[0006]

A document search device according to the present invention stores a document data storage means (18) for storing document data to be searched and a dictionary holding information for obtaining related words and word categories. Means (14), a judgment means (15) for judging whether the search keyword is classified into a plurality of categories by referring to the dictionary means, and a search keyword are classified into a plurality of categories by the judgment means. If it is determined that one or more words classified into each of the plurality of categories are taken out and presented, and a user specifies the designation means (16), the determination means determines a search keyword. When it is determined that the search keyword is classified into a plurality of categories, the determination unit determines that the search keyword is a single keyword based on the category of the word designated by the designation unit. When it is determined that the search keyword is classified, the keyword expansion means (13) for expanding the search keyword into a plurality of related words by referring to the dictionary means based on the category of the search keyword, and the search Searching means (17) for searching the document data with a keyword for a keyword and a word expanded by the keyword expanding means
And characterized in that: Further, according to an aspect of the present invention, the above-described document search device includes an input unit that inputs a search request in a sentence format, and a search keyword extraction unit that extracts a search keyword from the search request sentence.

[0007]

The dictionary means holds information for obtaining a related word and a word category for a word. For example, each headword is associated with information indicating the category and has the same category. Headwords constitute words that are related to each other. One entry word may correspond to a plurality of categories. The determination means refers to the dictionary means and determines whether or not the search keyword is classified into a plurality of categories. In the above example of the dictionary means, the determination means makes a determination by searching for an entry word that matches the search keyword and checking whether there are a plurality of categories corresponding to the matching entry word. When the determination unit determines that the search keyword is classified into a plurality of categories, the designation unit classifies the search keyword into each of the plurality of categories 1
Get more than one word, present it, and let the user specify it.
The keyword expanding means refers to the dictionary means in the category of the designated word and takes out the word belonging to the category. The search means searches the document data with the search keyword and the word extracted by the keyword expansion means. When the determination unit determines that the search keyword is classified into only one category, the keyword expansion unit refers to the dictionary unit to obtain the category of the search keyword, and finds words belonging to the category. Take out.
The search means searches the document data with the search keyword and the word extracted by the keyword expansion means.

When the search keyword is input, it may be directly input, but it is input in the form of a sentence requesting a search, and the keyword extracting means analyzes the sentence to obtain the search keyword. Can be configured.

[0009]

FIG. 1 is a functional block diagram showing a schematic configuration of an embodiment of a document search device according to the present invention. This document retrieval device is composed of an input unit 11, a keyword extraction unit 12, a keyword expansion unit 13, a dictionary unit 14, a category information determination unit 15, a keyword designation unit 16, a retrieval unit 17, a document data storage unit 18, and a display unit 19. Has been done.
The input unit 11 is a unit for inputting a search request. The search request is entered in the form of a search formula or question text. The keyword extracting unit 12 extracts a keyword serving as a search key from the search request input from the input unit 11. The keyword expansion means 13 uses the dictionary means 14 for the extracted keywords.
And expand the search key. The dictionary means 14 is shown in FIG.
As shown in, the headword 41 and the category information 42 corresponding thereto are stored. Category information determination means 15
When the keywords extracted by the keyword extracting means 12 are expanded by the keyword expanding means 13 using the dictionary means 14, it is determined whether or not each keyword belongs to a plurality of categories, and a plurality of categories are judged. If it belongs to, the word different from the keyword is taken out from each category and presented. Keyword designating means 16
Determines which of the words presented by the category information judging means 15 has been designated. Search means 17
Then, if the keywords extracted by the keyword extracting unit 12 are expanded by the keyword expanding unit 13 from the keywords input from the input unit 11 and the category information judging unit 15 judges that they belong to a plurality of categories, The document data storage unit 18 is searched using the key expanded by specifying the category to be expanded by the keyword specifying unit 16 as a search key. The document data storage unit 18 stores document data to be searched. The display means 19 displays the document data retrieved by the retrieval means 17.

The operation of the embodiment configured as described above will be described with reference to FIGS. FIG. 2 is a flow showing a search procedure in this embodiment. First, a search request including a search expression including a search keyword or a question sentence is input from the input unit 11 (step S21). This search formula and the like are sent to the keyword extracting means 12. The keyword extraction unit 12 performs morphological analysis on the search expression and extracts nouns and the like as keywords (step S22). When the keyword extracted by the keyword extracting unit 12 is sent, the keyword expanding unit 13 refers to the dictionary unit 14 to obtain category information corresponding to the keyword (step S23). As shown in FIG. 4, the dictionary means 14 includes a headword 41 and a category number 42 corresponding thereto, and when the headword is classified into a plurality of categories,
There are multiple category numbers. In the example of FIG. 4, a plurality of category numbers 11122, 11 are provided for the headword “teacher”.
233 and 11344 correspond. It is possible to obtain the category number from the entry word that matches the keyword and check whether or not the category number is classified into a plurality of categories.

The category information judging means 15 checks the category information obtained by the keyword expanding means 13 and judges whether the keyword is classified into a plurality of categories (step S24). When classified into a plurality of categories, one word is extracted from each category and displayed (step S25). That is, the dictionary means 14 of FIG. 4 is referred to for each of the obtained plurality of category numbers, and each one of the corresponding headwords is obtained. It is displayed as shown in FIG. The user sees the word that is determined and displayed by the category information determination means 15 to be classified into a plurality of categories, and selects a word close to the search intention. The keyword designating means 16 accepts the selection of the user and informs the keyword expanding means 13 (step S2).
6).

The keyword expanding means 13 compares the category number of the category corresponding to the selected word with the category number 42 of the dictionary means 14, and extracts the entry words of all the matching entries. (Step S27). This gives headwords that belong to the same category as the selected word, and these constitute a group of related words. As a result of the determination in step S24, the related word of each keyword is 1
If it is classified into only one category, the category number of that category is compared with the category number 42 of the dictionary means 14, and the key words are expanded by extracting the entry words of all the matching entries (step S28). The search means 17 searches the document data storage means 18 with the word expanded by the keyword expansion means 13 as a search key. The display unit 19 displays the document data retrieved by the retrieval unit 17 (step S29).

Next, an example will be described in which a search expression "ask a question to the teacher" is actually given as a search request. When "Ask the teacher a question" is input from the input unit 11 as shown in FIG. 3A, the keyword extraction unit 12 extracts a noun by a known morphological analysis,
As in (b), "teacher" and "question" are obtained as keywords. The extracted keywords are compared by the keyword expansion means 13 with the entry words of the dictionary means 14 of FIG.
Extract the category number of the matching entry. In the case of the example in FIG. 4, the category number 1 is applied to the keyword “teacher”.
1122, 11233, 11344 are obtained, and the category number 12323 is obtained for the keyword “question”. The category information judging means 15 checks whether or not there are a plurality of category numbers obtained as described above, and when there are a plurality of category numbers, that is, in the case of “teacher” in the above example, the category means 14 determines from the corresponding category number. One headword corresponding to each category number is extracted with reference to FIG.
Display as in (a). “Teacher” is displayed for the category number 11122, “Teacher” is displayed for the category number 11233, and category number 11344 is displayed.
Is displayed as "doctor". The keyword designating means 16 allows the user to select a word close to the search intention from the displayed words. FIG. 5B shows a state where “teacher” is selected. This selection information is given to the keyword expanding means 13. Keyword expansion means 13
Uses the category number of the selected word "teacher" to look up the dictionary means 14 to obtain a headword having the same category number. Since the category number of "teacher" is 11122, the headwords "teacher", "teacher", and "professor" having the category number are displayed together with the original keyword "teacher".

On the other hand, since the "question" extracted by the keyword extracting means 12 has only one category number,
The category information determination means 15 determines that the category information is not classified into a plurality of categories, and upon receiving the determination, the keyword expansion means 13 immediately expands the keyword. "Question"
When the dictionary means 14 is referred to by the category number 12323, the "question" and "question" having the same category number can be obtained. In this way, the search key obtained by the expansion is displayed as shown in FIG.

In the search by the search means 17, the document data including any of the keywords expanded by "teacher" and the document data including any of the keywords expanded by "question" are searched for, and the common data is obtained. The document data of is used as the search result. The file name of the retrieved result is displayed as shown in FIG.

In the present embodiment, a sentence is input as a search expression, but a word serving as a search key may be directly input. In that case, the keyword extracting means 12 of FIG. 1 is configured as shown in FIG. The output of the input unit 11 is directly connected to the input of the keyword expansion unit 13, and a similar search can be realized.

Further, as shown in FIG. 7A, the dictionary means 14
If information indicating whether or not a word corresponding to the dictionary entry word exists in the document data to be searched is added to the dictionary contents of, the keyword that does not exist in the document is not expanded as a search key. Therefore, the search efficiency can be improved. In FIG. 7A, the information about the existence or non-existence is shown as a document flag. When the information exists in the document data, the document flag is set to "1", and otherwise, "0" is set. FIG. 7B shows a search key expanded for the keyword existing in the document data. Here, from the contrast between FIG. 5C and FIG. 7B in which all the words in the specified category are expanded as the search key,
It can be seen that words that do not exist in the document are omitted from the search key.

Further, when the category is designated by the keyword designating means 16, the designation of the category is not limited to one, but a plurality of categories can be designated. further,
In displaying, not only the list of document names but also the contents of the document at the top of the list may be displayed together with the list, and the display method is not particularly limited as in the present embodiment.

[0019]

As described above, in the document search device of the present invention, for words belonging to a plurality of categories,
Extract and display one or more words from each category. This allows the user to easily determine which category is suitable for the search intention even for keywords having a complicated semantic relationship that cannot be represented by a tree structure. Then, by allowing the user to select a word that is close to the search intention, words that do not match the search intention that are classified into other unselected classifications are not expanded, so that the desired search result can be obtained efficiently.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing a schematic configuration of an embodiment of a document search device of the present invention.

FIG. 2 is a flowchart showing the operation of this embodiment.

3A is a diagram showing a search request input from an input unit, FIG. 3B is a diagram showing an example of a keyword extracted by a keyword extracting unit, and FIG. 3C is an example of expanding a keyword by a dictionary. FIG.

FIG. 4 is a diagram showing an example of contents of a dictionary.

FIG. 5A is a diagram showing an example in which words are extracted from each category for keywords classified into a plurality of categories, and FIG. 5B is a diagram showing words close to a search intention from the words extracted from each category. The figure which shows one example which was designated, (c)
Is an explanatory diagram showing an example of actually expanded keywords,
(D) The figure which shows an example of a display of a search result.

FIG. 6 is a functional block diagram showing a schematic configuration of another embodiment.

FIG. 7A is a diagram showing an example in which information indicating whether or not a document exists is added to a dictionary, and FIG. 7B is a diagram showing an example in which only keywords existing in a document are expanded.

[Explanation of symbols]

11 ... Inputting means, 12 ... Keyword extracting means, 13 ... Keyword expanding means, 14 ... Dictionary means, 15 ... Category information judging means, 16 ... Keyword specifying means, 17 ... Searching means, 18 ... Document data storing means, 19 ... Display means.

(72) Inventor Akio Yamashita 3-2-1 Sakado, Takatsu-ku, Kawasaki-shi, Kanagawa KSP R & D Business Park Building Fuji Xerox Co., Ltd. (72) Kazuo Aihara 3-chome, Sakado, Takatsu-ku, Kawasaki-shi, Kanagawa 2-1 KSP R & D Business Park Building, Fuji Xerox Co., Ltd. (72) Inventor Tatsuomi Kita 3-2-1 Sakado, Takatsu-ku, Kawasaki City, Kanagawa KSP R & D Business Park Building, Fuji Xerox Co., Ltd. (72) Inventor Yuko Matsuo 3-2-1 Sakado, Takatsu-ku, Kawasaki-shi, Kanagawa KSP R & D Business Park Building Fuji Xerox Co., Ltd. (72) Inventor Shinji Kawamoto 3-2-1, Sakado, Takatsu-ku, Kawasaki-shi, Kanagawa KSP R & D Business Park Building Tomi Shishi Xerox Co., Ltd.

Claims (2)

[Claims] 1. A document data storage unit that stores document data to be searched, a dictionary unit that holds information for obtaining a related word and a category of the word, and a dictionary for referring to the dictionary unit. Determination means for determining whether the search keyword is classified into a plurality of categories, and one piece classified from each of the plurality of categories when the search keyword is determined to be classified into a plurality of categories The above-mentioned words are taken out and presented, and a designating unit that allows the user to designate, and when the determining unit determines that the search keyword is classified into a plurality of categories, The category of the search keyword is based on the category, and when the determination means determines that the search keyword is classified into one category. And a keyword expanding means for expanding the search keyword into a plurality of related words by referring to the dictionary means, and a search for searching the document data with the search keyword and the word expanded by the keyword expanding means. A document retrieval device comprising means. 2. The document according to claim 1, further comprising input means for inputting a search request in the form of a sentence, and search keyword extracting means for extracting a search keyword from the sentence of the search request. Search device.